Retrieving formula for "Mean Confidence Interval" from Fit Model

[#1]

Hello,

After doing a fit model, I can easily save a new column for the prediction formula by going:

Red Triangle -> Save Columns -> Prediction Formula

I would also like to save the formula for the Mean Confidence Interval, but it seems like that is not available.

I can generate static columns for the Mean Confidence Interval by going:

Red Triangle -> Save Columns -> Mean Confidence Interval

Is there a way to get the formula used to generate these values, or any kind of workaround known that will eventually allow me to find this formula?

Thank you,

-Tom

[#2]

In the same Save Columns menu, do you see the StdErr Pred Formula and the Mean Confidene Limit Formula commands? These commands should help.

What version of JMP are you using?

[#3]

Hi Mark,

I have StdErr Prediction Formula, but not Mean Confidence Limit Formula.

I'm using JMP non-Pro 12.1.0. I could get 13 installed if necessary, although it's not officially "approved software" where I work yet. Would this be in 13?

[#4 · Accepted solution]

I cannot check JMP 12 at the moment. I also cannot keep track of the features in every version of JMP! But I see it in JMP 13. Maybe another JMP staffer will pipe up to confirm if JMP 13 is necessary!

In the mean time, though, it is easy to calculate yourself. Follow these steps:

1. Fit your model.
2. Save the Prediction Formula.
3. Save the StdErr Prediction Formula.
4. Examine the Analysis of Variance table and note the Error DF.
5. Add a new column to the data table for the lower limit.
6. Add a formula to this column.
   1. Select the column with the Prediction Formula and click the Subtract button.
   2. Select Probability > t Quantile.
   3. Enter 0.975 for the p argument.
   4. Enter the error DF you noted for the DF argument.
   5. Select the t Quantle function in your formula and click the Multiply button.
   6. Select the StdErr Prediction Formula column.
   7. Click OK.

Now, do the same for the upper bound, just add instead of subtract. (I assumed that you wanted two-sided limits.)

[#5]

This is perfect, thanks!

[#6]

Also: I just discovered that "Mean Confidence Limit Formula" does show up in JMP 12 as long as the model does not contain Random effects.

[#7]

Hi, I just read this comment and was wondering how the confidence intervals would be calculated in case of a mixed model (with one random effect)?

[#8]

Using JMP Pro 12.1.0 on a Win 7 pc I do see the option to save the mean confidence interval formula. I hope this doesn't mean that the Pro version is required to get the mean confidence interval formula!